INTRODUCTION — The term acute myeloid leukemia (AML) refers to a group of hematopoietic neoplasms involving cells committed to the myeloid line of cellular development. AML is characterized by a clonal proliferation of myeloid precursors with reduced capacity to differentiate into more mature cellular elements.
The response to treatment and overall survival of patients with AML is heterogeneous. A number of prognostic factors related to patient and tumor characteristics have been described for AML, including age, performance status, and karyotype (table 1) [1-3].
This topic will review prognostic factors in AML. Adverse risk factors that are more common in older adults (eg, patients over age 60 years) with AML are discussed separately. (See "Pretreatment evaluation and prognosis of acute myeloid leukemia in older adults", section on 'Prognosis'.)
The diagnosis, treatment, complications of AML, and further information on specific cytogenetic abnormalities are also discussed separately. (See "Clinical manifestations, pathologic features, and diagnosis of acute myeloid leukemia" and "Cytogenetic abnormalities in acute myeloid leukemia" and "Induction therapy for acute myeloid leukemia in medically-fit adults" and "Acute myeloid leukemia: Management of medically-unfit adults" and "Overview of the complications of acute myeloid leukemia".)
CLINICAL RISK FACTORS — There are several clinical findings that may help predict the likelihood of attaining a complete remission (CR) and subsequent disease-free survival (DFS) in patients with AML. The strongest adverse clinical predictors are:
●Advanced age
●Poor performance status
●Cytogenetic and/or molecular genetic findings in tumor cells
●History of prior exposure to cytotoxic agents or radiation therapy
●History of prior myelodysplasia or other hematologic disorders such as myeloproliferative neoplasms
The first two factors are the main predictors of early death, while the others are better predictors of resistant disease or early relapse. (See "Therapy-related myeloid neoplasms: Epidemiology, causes, evaluation, and diagnosis" and 'Karyotype' below.)
Age — While there is no clearly accepted definition of younger compared to older adults when dealing with AML, in most studies, "older adults" was defined as greater than ages 55, 60, or 65 years. Such older adults have lower rates of achieving a CR and shorter DFS when compared with younger patients.
Higher age appears to act as an adverse prognostic marker even in younger patients. A population-based retrospective study from the United Kingdom that included 11,303 patients with AML diagnosed from 2001 to 2006 reported an estimated five-year overall survival (OS) rate of 15 percent, which varied according to the age at diagnosis [4]:
●15 to 24 years (289 patients) – 53 percent
●25 to 39 years (702 patients) – 49 percent
●40 to 59 years (2170 patients) – 33 percent
●60 to 69 years (2208 patients) – 13 percent
●70 to 79 years (3258 patients) – 3 percent
●>80 years (2676 patients) – 0 percent
A retrospective analysis of 891 patients <30 years old found that five-year event-free survival (EFS) rates decreased as patient age at diagnosis increased with rates of 54, 46, and 28 percent for children (2 to <13 years), adolescents (13 to <21 years), and young adults (21 to <30 years), respectively [5]. With favorable karyotypes excluded, outcomes were superior in children (EFS 44 percent) compared with adolescents (35 percent) and young adults (23 percent).
The effect of age on outcome in patients with AML is discussed in more detail separately. (See "Pretreatment evaluation and prognosis of acute myeloid leukemia in older adults", section on 'Prognosis'.)
Performance status — Most other clinical risk factors in AML are related to comorbidities (eg, heart failure, renal insufficiency, concurrent infection) that can be partially reflected by the patient's performance status [3,6]. The two most commonly used tools are the Karnofsky performance status and the Eastern Cooperative Oncology Group (ECOG) performance status (table 2A-B). Performance status and age at diagnosis can be combined to estimate the percentage of patients who will die within the first 28 days of treatment [3]. This ranges from 5 percent for patients under the age of 50 years with an ECOG performance status <3 to 57 percent for patients over age 69 with an ECOG performance status of ≥3 (table 3). Performance status appears to be of greatest prognostic value in older adults and may not predict early outcomes (mortality, intensive care unit admission, CR) in younger patients [7]. (See "Pretreatment evaluation and prognosis of acute myeloid leukemia in older adults", section on 'Physical functioning'.)
As an example, a retrospective analysis of 968 adults with newly diagnosed AML treated by the Southwest Oncology Group (SWOG) enrolled in prospective trials of induction therapy reported increasing 30-day mortality rate with worsening performance status and increased age at diagnosis [8]. Patients ≤65 years of age had 30-day mortality rates of 5, 4, 9, and 21 percent if they presented with an ECOG performance status of 0, 1, 2, or 3, respectively. Patients >65 years had rates of 13, 16, 35, and 60 percent, respectively. Older patients with a worse performance status were also less likely to obtain a CR. However, subsequent studies have reported lower rates of short-term mortality in older patients, probably due to improvements in supportive care. As an example, a large randomized trial in patients >60 years of age reported a 30-day mortality of approximately 11 percent [9].
It should be emphasized that these and other published estimates of outcomes derive from results of patients entered on clinical trials. These trials usually exclude individuals with organ dysfunction and substantially impaired performance status and hence these results may not directly apply to the large fraction of older patients with medical and sometimes psychosocial comorbidities. Indeed, data indicate that the majority of older patients do not receive induction chemotherapy as a consequence of either their refusal of treatment or physician assessment that they would not tolerate or benefit from treatment. Assessment of performance status in AML can be tricky. For example, a patient may have no or minimal symptoms with a PS of 0 or 1 and develop fever and infection and instantly become a PS 3 or 4, depending on the severity of the infection. With proper treatment, they can rapidly revert to baseline. Therefore, many factors may have to be considered when using PS to predict prognosis, and it may be that medical comorbidity indices that also include more chronic medical conditions will be more helpful in the future. A number of such scoring systems, some originally designed for transplant recipients, have been developed, but none have been validated for use in individual patients.
Therapy-related AML — Persons who are exposed to cytotoxic agents or radiation therapy are at risk of developing AML, myelodysplastic syndrome (MDS), and myelodysplastic syndrome/myeloproliferative neoplasms (MDS/MPN). These conditions lie along a continuum of disease and are categorized by the 2016 World Health Organization classification system as therapy-related myeloid neoplasms (t-MN) [10]. T-MNs account for approximately 10 to 20 percent of all cases of AML, MDS, and MDS/MPN. The incidence among patients treated with cytotoxic agents varies according to the underlying disease, specific agents, timing of exposure, and dose. The prognosis of patients with t-MN is generally worse than for those with de novo AML because of higher rates of drug resistance. This is discussed in more detail separately. (See "Therapy-related myeloid neoplasms: Epidemiology, causes, evaluation, and diagnosis" and "Cytogenetic abnormalities in acute myeloid leukemia", section on 'Therapy-related myeloid neoplasms'.)
Antecedent hematologic disorders — Pre-existing myelodysplastic or myeloproliferative disorders are common in older patients with AML, occurring in 24 to 40 percent of cases [11-15]. These disorders are often associated with ineffective hematopoiesis and dysfunctional blood cells. By the time that AML emerges, these patients may be colonized by pathogenic flora, threatened by recurrent bleeding episodes, and dependent on transfusions.
In one study that compared gene mutations in patients with secondary AML (after a prior myeloid malignancy), therapy-related AML, and unspecified AML, the identification of a gene mutation in one of eight genes (SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR, or STAG2) was highly specific for a diagnosis of secondary AML and a poor clinical outcome [16].
Other factors — The effect of race on patient outcomes in AML is uncertain. Using data from seven CALGB studies, it was found that African American and White subjects with AML differ significantly with respect to important risk factors such as cytogenetic risk groups and age [1]. African-American men had significantly lower rates of CR (54 versus 64 percent) and five-year OS (16 versus 24 percent) than White men, and may represent a higher risk group. Further studies of this type are needed to confirm these results, suggest possible biologic mechanisms, and provide appropriate high-risk treatment programs.
CYTOGENETIC AND MOLECULAR FEATURES — Specific cytogenetic and molecular features permit stratification of AML into various prognostic groups. Understanding of the impact of combinations of cytogenetic and molecular findings is evolving. We prefer an approach similar to the classification system proposed by the European LeukemiaNet (ELN) (table 4) [17]. Treatment decisions that are informed by risk stratification are discussed separately. (See "Post-remission therapy for acute myeloid leukemia in younger adults" and "Treatment of relapsed or refractory acute myeloid leukemia".)
Information regarding the genetic landscape of AML and its impact on prognosis comes from an analysis of 1540 patients with AML enrolled on prospective trials of intensive therapy that included correlative studies with cytogenetic analysis and gene sequencing [18]. This study identified 14 major genomic subgroups of AML that largely correspond to the ELN classification.
European LeukemiaNet classification — The ELN integrates cytogenetic and molecular features (eg, FLT3-ITD, CEBPA, and NPM1) in AML to divide cases into three prognostic risk groups that differ based on rates of complete remission (CR), disease-free survival (DFS), and overall survival (OS) (table 4) [17]:
●Favorable risk:
•t(8;21)(q22;q22.1); RUNX1-RUNX1T1
•inv(16)(p13.1;q22) or t(16;16)(p13.1;q22); CBFB-MYH11
•Mutated NPM1 without FLT3-ITD or with low allelic ratio (<0.5) of FLT3-ITD
•Biallelic mutated CEBPA
●Intermediate risk:
•Mutated NPM1 and high allelic ratio (>0.5) of FLT3-ITD
•Wild-type NPM1 without FLT3-ITD or with low allelic ratio (<0.5) of FLT3-ITD (without adverse-risk genetic lesions)
•t(9;11)(p21.3;q23.3); MLLT3-KMT2A
•Cytogenetic abnormalities not classified as favorable or adverse
●Adverse risk:
•t(6;9)(p23;q34.1); DEK-NUP214
•t(v;11q23.3); KMT2A rearranged
•t(9;22)(q34.1;q11.2); BCR-ABL1
•inv(3)(q21.3;q26.2) or t(3;3)(q21.3;q26.2); GATA2, MECOM (EVI1)
•Monosomy 5 or del(5q); monosomy 7; monosomy 17/abn(17p)
•Complex karyotype, monosomal karyotype
•Wild-type NPM1 and high allelic ratio (>0.5) of FLT3-ITD
•Mutant RUNX1, ASXL1, or TP53
Note the use of "allelic ratio" to stratify based on the FLT3-ITD mutation. Various research groups calculate this ratio differently and/or use other cutoff values. This calculation is affected by the admixture of remaining normal bone marrow cells, which add normal alleles to the denominator, thereby "diluting" or reducing the calculated ratio. Thus, this measure may only be reliable if the majority of the cells are leukemic and the fraction is between 0.5 and 1.0, which would necessarily indicate that the malignant cells are either homozygous or hemizygous for the FLT3 mutant allele.
A "monosomal karyotype" is defined as at least two autosomal monosomies or a single autosomal monosomy in the presence of one or more structural cytogenetic abnormalities. Additional details regarding the stratification scheme are provided in the table (table 4) [17].
Other models include clinical features in addition to cytogenetic and molecular features. The prognostic index for cytogenetically normal AML (PINA) combines cytogenetic and molecular data with patient age, performance status, and white blood cell count at diagnosis to identify three distinct cohorts (low, intermediate, and high risk) with statistically different estimated rates of OS (74, 28, and 3 percent) and cumulative incidence of relapse (35, 56, and 72 percent) at five years [19].
Karyotype
General — Karyotype analysis with metaphase cytogenetics is a key component of the initial evaluation of a patient with AML; specific cytogenetic abnormalities in AML have considerable prognostic significance and affect treatment planning (table 5 and table 4). The value of risk stratification by karyotype has been illustrated in several analyses of patients enrolled in prospective clinical trials. The largest studies were cooperative group efforts from the Medical Research Council (MRC), the Southwest Oncology Group/Eastern Cooperative Oncology Group (SWOG/ECOG), and the Cancer and Leukemia Group B (CALGB). All studies confirmed earlier results from other groups attesting to the importance of pretreatment karyotype (figure 1) [20-27]. The following describes methods of grouping karyotype abnormalities by risk of progression. Details regarding specific chromosomal changes and their impact on prognosis are presented separately. (See "Cytogenetic abnormalities in acute myeloid leukemia".)
The specifics regarding what constitutes favorable, intermediate, and unfavorable risk have varied among the cooperative groups. While there has been general agreement that t(8;21), inv(16), and t(15;17) predict a good outcome, there has been disagreement regarding what abnormalities determine an unfavorable risk and how additional chromosomal abnormalities impact the prognostic value of known markers.
Information from 5876 adults with newly diagnosed de novo (93 percent) or secondary AML enrolled on prospective MRC trials of intensive anthracycline- and cytarabine-based combination chemotherapy [27] was used to modify the MRC's prior stratification system [28,29] to create the risk stratification system shown below. Using these definitions, rates of OS at 10 years were 69, 38, 33, and 12 percent for patients with favorable risk, normal karyotype, intermediate risk, and adverse risk, respectively.
●Favorable (approximately 16 percent of newly diagnosed patients) – The following abnormalities are considered favorable, whether alone or in conjunction with other abnormalities: t(8;21), inv(16)(p13;q22), t(16;16)(p13;q22). t(15;17)(q24.1;q21.1) which identifies acute promyelocytic leukemia (APL) is also favorable, but APL is now considered separately from other forms of AML because different treatment approaches are used.
●Normal karyotype (approximately 40 percent)
●Intermediate (approximately 20 percent) – Abnormalities not described in favorable or unfavorable.
●Adverse or poor (approximately 25 percent) – The following abnormalities are considered unfavorable when they occur in cases that do not also contain favorable karyotypic changes: del (5q); add (5q); del (7q); add (7q); monosomies 5 or 7; inv(3)(q21q26); t(3;3)(q21;q26); t(6;11)(q27;q23); t(10;11)(p11-13;q23); t(9;22)(q34;q11); 17p abnormalities or monosomy 17; complex aberrant karyotypes described as at least 4 unrelated abnormalities; 11q23 abnormalities excluding t(9;11)(p21;q23) and excluding t(11;19)(q23;p13); or abnormalities of 3q excluding t(3;5)(q21-25;q31-35).
These survival rates are consistent with those seen by other cooperative groups using similar risk stratification systems [22,30]. This study also served to clarify the prognostic value of findings that were previously not well understood. As examples:
●A "monosomal karyotype," defined as at least two autosomal monosomies or a single autosomal monosomy in the presence of one or more structural cytogenetic abnormalities, has been proposed as a better predictor of unfavorable-risk disease than a complex karyotype [31]. The approximately 10 percent of patients with AML who demonstrate a monosomal karyotype have a low rate of CR after induction (48 percent) and a <5 percent rate of OS at four years [31,32]. The additional prognostic value of including "monosomal karyotype" in the adverse risk group was specifically addressed in the analysis of 5876 patients on MRC trials [27]. While patients with a monosomal karyotype had particularly poor outcomes (<5 percent OS at 10 years), 94 percent of cases with a monosomal karyotype were already classified as adverse risk according to the criteria defined above. In at least one retrospective study, the prognostic values of the monosomal karyotype and complex karyotype did not remain when patients with 17p abnormalities or chromosome 5 abnormalities were removed from the analysis [33].
In two large retrospective analyses, patients with a monosomal karyotype had lower rates of CR and OS when compared with patients with unfavorable cytogenetics without a monosomal karyotype [34,35]. The overall outcome for patients with a monosomal karyotype appeared to be better following allogeneic hematopoietic cell transplantation (HCT) when compared with consolidation with chemotherapy alone. The benefit of allogeneic HCT in this population was further supported by an analysis of 305 patients with newly diagnosed AML with a monosomal karyotype enrolled in three consecutive HOVON/SAKK phase III trials [36]. CR was achieved in 140 (46 percent) and 107 proceeded with consolidation chemotherapy (48 patients), high dose chemotherapy with autologous stem cell rescue (14 patients), or allogeneic HCT (45 patients). The rates of overall and relapse-free survival (RFS) at five years were 13 and 12 percent, respectively. The five-year OS rates were higher with allogeneic HCT when compared with chemotherapy or autologous stem cell rescue (19 versus 9 percent). None of the 33 patients who achieved a CR but did not proceed to consolidation survived.
●Patients with abnormalities of chromosome band 11q23, seen in 5 to 10 percent of patients with de novo AML and in some patients with therapy-related leukemia after having received topoisomerase II inhibitors, have had a poor outcome when treated with conventional chemotherapy and have previously been included in the adverse risk group [37-40]. Data from the MRC analysis supported prior suggestions that a subset of patients with t(9;11) may be cured with intensive consolidation chemotherapy alone (10 year OS rate of 39 percent) [27,41,42]. As such, patients with this particular 11q23 translocation are not considered to have adverse risk based on this abnormality.
●Prior to the MRC analysis, it was not clear how to categorize patients with isolated trisomy 8 with some studies suggesting that these patients had adverse outcomes [43] and others reporting that their outcomes did not differ from those with normal cytogenetics [44]. The 547 patients with trisomy 8 included in the MRC database had a similar prognosis to those with normal cytogenetics (10-year OS in 37 percent) [27].
While patients with cytogenetically normal AML have traditionally been included in the intermediate risk group category, more sophisticated analyses, including gene mutation and microRNA expression studies, suggest that this group is more heterogeneous than previously thought. The effect of gene mutations and microRNA expression on prognosis is discussed in more detail below. (See 'Gene mutations' below and 'MicroRNA expression profiling' below.)
Older adults — Similar stratification systems to that used in the general population (described above) have also shown prognostic value in older adults with AML [11,13,45-48]. The more favorable translocations seen with core-binding factor AML are considerably less common in older adults. In this population, patients with favorable (7 percent), intermediate (73 percent), or poor risk (20 percent) cytogenetics had CR rates of 72, 53 to 63, and 26 percent, respectively [45]. OS rates at five years for the same groups were 34, 10 to 15, and 2 percent, respectively.
●A study from the CALGB included 42 patients with isolated trisomy 8, 60 percent of whom were more than 60 years of age [43]. Median survival was lower in trisomy 8 patients over the age of 60 than in younger patients (4.8 versus 17.5 months) with no long-term survivors among the older patients. The only long-term survivors were those less than 60 years of age who were treated with autologous or allogeneic HCT while in first CR (CR1). A later CALGB study indicated a five-year OS of 7 to 9 percent for the group of patients >60 years of age with trisomy 8 [49].
●In another CALGB study in patients ≥60 years of age with AML, the presence of a complex karyotype consisting of three or more chromosome abnormalities in at least one clone or a rare aberration (in the absence of t(8;21) and inv(16)), was associated with five-year DFS and OS rates of 2 percent or less following standard cytotoxic chemotherapy used in conventional doses [49]. Thus, such patients are better suited for investigational therapy on a clinical trial or supportive care. Similar conclusions were made in a German-Austrian study, in which three-year OS for patients >70 years of age with high-risk cytogenetics was 2 percent [47].
Gene mutations — AML is typically categorized based on combinations of mutations and/or altered expression of certain genes (table 4). Many centers routinely perform genetic profiling on all newly diagnosed patients. Abnormalities in certain genes (eg, mutations in FLT3, NPM1, KIT) and altered gene expression profiles confer prognostic significance in adult patients with AML [50]. This is particularly important in the approximately 45 percent of patients with normal karyotypes. This heterogeneous group of patients contains some with a better prognosis (eg, mutant CEBPA) and others with an adverse prognosis (eg, FLT3-ITD) (table 1 and table 6) [51-67].
Abnormalities in FLT3, NPM1, KIT, CEBPA, TP53, RUNX1, and ASXL1 have been the most widely studied. Other gene mutations, such as those involving WT1 (Wilms tumor 1), meningioma 1 (MN1), TET2, IDH1, IDH2, DNMT3A, SRSF2, or RAS may also have prognostic significance but need further confirmation in prospective studies [68-92]. (See "Molecular genetics of acute myeloid leukemia", section on 'Mutations affecting DNA methylation'.)
As an example, one report described a panel of 17 genes that reflected the functional leukemic stem cell phenotype of specimens from newly diagnosed AML [93]. The clinical impact of many of the mutations identified by these tests is uncertain, and only a few of these abnormalities are currently "druggable" or targeted by available agents. Hopefully, better understanding of the mechanisms by which these changes produce increased or decreased sensitivity to treatment will evolve, permitting more rational choices of chemotherapy and the selective application of transplantation.
FLT3 gene — FLT3 is a transmembrane tyrosine kinase receptor that stimulates cell proliferation upon activation. Mutations in the FMS-like tyrosine kinase 3 gene producing internal tandem duplications (FLT3-ITD) and constitutive activation of the FLT3 receptor tyrosine kinase are quite common in AML, particularly in patients with normal karyotypes, and have been associated with poorer OS in children and in younger and older adults receiving intensive chemotherapy [56,82,94-105]. It has been proposed that FLT3-ITD mutational status is the primary predictor of outcome among patients with intermediate-risk AML by karyotype analysis [104]. The exact frequency varies with age with FLT3 mutations being present in approximately 10 and 30 percent of patients with cytogenetically normal AML in the pediatric and adult populations, respectively [100,106,107]. Concurrent abnormalities in other genes, such as NPM1, may influence the impact of the FLT3 mutation [108].
There are two main types of FLT3 mutations. The most common are internal tandem duplications (ITD) of different length that result in ligand-independent activation of the FLT3 receptor and a proliferative signal. Additionally, in 5 to 10 percent of patients, point mutations in the activating loop of the kinase domain of FLT3 may also result in activation of FLT3 [109,110]. The prognostic impact of FLT3-ITD is influenced by its mutational context, including the absence of the wild-type FLT3 allele (ie, homozygous or hemizygous FLT3-ITD), the concurrent mutation status of NPM1, and the FLT3 mutant/wild-type allelic ratio (AR). Homozygous or hemizygous FLT3 and higher AR are associated with poorer outcomes [96,111,112]. Patients with mutant NPM1 and FLT3-ITD with a low AR (<0.5) have more favorable outcomes, whereas patients with wild-type NPM1 and FLT3 AR ≥0.5 have poor outcomes [101,113,114].
The MRC conducted an analysis of the results of both autologous and allogeneic HCT in patients whose leukemia cells were positive for FLT3-ITD [115]. This was a retrospective evaluation of large studies of patients who had been randomly assigned to these HCT approaches irrespective of their FLT3 mutational status. In FLT3-ITD positive patients, there was no difference in RFS or OS between patients allocated to either type of HCT and those treated with chemotherapy alone. However, somewhat conflicting results were seen in similar retrospective analyses conducted by German leukemia groups who noted improved outcomes in FLT3 mutated patients who had a histocompatible sibling donor compared with those who did not [101].
In another study, the European Group for Blood and Marrow Transplantation (EBMT) analyzed the outcomes of 206 patients with AML who underwent HLA-identical sibling or matched unrelated allogeneic HCT in CR1 after myeloablative conditioning [116]. The 120 patients with FLT3-ITD had a higher median leukocyte count at diagnosis and a shorter interval from CR to HCT, but otherwise had similar baseline characteristics to those without FLT3-ITD. The presence of FLT3-ITD was associated with a higher estimated cumulative incidence of relapse at two years post-transplant (30 versus 16 percent), and a lower two-year leukemia-free survival rate (58 versus 71 percent).
Use of midostaurin (FLT3 inhibitor) as a component of remission induction therapy and the importance of FLT3 status on selection of post-remission therapy for AML are discussed separately. (See "Induction therapy for acute myeloid leukemia in medically-fit adults", section on 'Inhibitors of mutated FLT3' and "Post-remission therapy for acute myeloid leukemia in younger adults", section on 'Unfavorable-risk disease'.)
NPM gene — Abnormalities in the nucleophosmin (NPM1) gene are found in approximately 25 and 50 percent of patients with de novo AML or de novo normal karyotype AML, respectively. NPM1 mutations have been associated with improved outcomes in younger and older adults, and children with AML, although the mechanism for increased chemosensitivity is not known [1,82,117-121]. However, concurrent cytogenetic abnormalities, mutations in other genes (eg, FLT3), and the NPM1 mutant allele burden influence the impact of the NPM1 mutation [104,122-124]. The superior prognosis is limited to those with NPM1 mutation who do not have a FLT3-ITD mutation and a normal karyotype.
●In 215 younger adults with newly diagnosed AML enrolled on prospective MRC trials, patients with a normal karyotype AML and FLT3-ITD with wild-type NPM1 had a poor prognosis (13 percent alive at 10 years) while patients with NPM1 mutation without FLT3-ITD demonstrated superior OS rates (50 percent alive at 10 years) [27]. There were not enough data regarding CEBPA status to analyze its effect on prognosis [101,125-128].
●Subjects with a normal karyotype, NPM1 mutations, wild-type FLT3, and low levels of ERG appear to have an especially favorable prognosis, with an estimated two-year PFS of 70 percent after induction treatment with cytarabine, daunorubicin, and etoposide followed by intensification with high dose cytarabine or intensification followed by autologous HCT [56]. In another study, the four-year OS in similar patients was approximately 60 percent [101].
●Pooled analysis from nine international studies identified 2426 patients with NPM1 mutations and FLT3-wild type or FLT3-ITD with low allelic burden, including 2000 with normal karyotype and 426 with abnormal karyotype [124]. In patients with this molecular profile, those with adverse cytogenetics had lower rates of CR and five-year OS and event-free survival (EFS), indicating that higher cytogenetic risk negates the favorable molecular risk of NPM1 mutations and FLT3-wild type or FLT3-ITD with low allelic burden. It should be noted that only 3 percent of the overall NPM1 mutated group had a high-risk karyotype.
Older patients with NPM1 mutations also have improved outcomes. As an example, a study from CALGB demonstrated high rates of CR and a three-year OS rate of 35 percent in patients whose blasts were NPM1 mutated and FLT3 wild type [120].
The impact of NPM1 status on selection of post-remission therapy for AML is discussed separately. (See "Post-remission therapy for acute myeloid leukemia in younger adults", section on 'Favorable-risk disease'.)
CEBPA gene — The CEBPA (CCAAT/enhancer binding protein alpha) gene encodes a transcription factor essential for myeloid differentiation [129-131]. CEBPA mutations are one of two known mutation types associated with familial leukemia and can be found in approximately 10 percent of patients with newly diagnosed AML [104,132]. In addition, 13 to 19 percent of patients with cytogenetically normal AML will have CEBPA mutations [52,133-137]. Familial AML with mutated CEBPA has a phenotype that is similar to sporadic AML with biallelic CEBPA mutations. Patients with cytogenetically normal AML with CEBPA mutations have a significantly longer median OS that is independent of other high-risk molecular features [82,132,133,138,139]. (See "Familial disorders of acute leukemia and myelodysplastic syndromes", section on 'Familial AML with mutated CEBPA'.)
As an example, a prospective analysis followed 175 adult patients with newly diagnosed AML with normal cytogenetics who were less than 60 years of age for a median of 4.8 years [132]. Patients with CEBPA mutations had significantly higher rates of five-year EFS (53 versus 30 percent) when compared with those with wild-type CEBPA. In a subset analysis, patients with high-risk molecular features (ie, FLT3-ITD positive and/or wild-type NPM1) who had mutations in CEBPA had significantly better rates of five-year EFS (55 versus 17 percent) and OS (58 versus 27 percent) when compared with those with unmutated CEBPA.
The favorable effect of CEBPA mutations may be limited to patients who carry two copies of the mutant allele and are negative for FLT3-ITD mutations and IDH2 R140 mutations [18,138,140-143]. An international study of 1182 patients with cytogenetically normal AML reported double mutations of CEBPA (either two different mutations or one homozygous mutation) in 91 cases and single mutations in 60 cases [137]. When compared with patients with double mutations of CEBPA, patients with a single CEBPA mutation had higher rates of concurrent mutations in NPM1 (35 versus 3 percent) and FLT3-ITD (30 versus 8 percent). While the presence of any CEBPA mutation was associated with a favorable outcome, only the presence of double mutations of CEBPA was an independent prognostic factor on multivariable analysis. Therefore, the data mentioned above, some of which included patients with only a single mutated allele, may need further study. In addition, there appears to be a separate group of patients in whom CEBPA expression is "silenced" by DNA hypermethylation, the prognostic effect of which is unknown [144].
IDH genes — Somatic mutations in the genes encoding isocitrate dehydrogenase (IDH1 and IDH2) are present in approximately 15 percent of newly diagnosed AML [60-65,104]. IDH1/2 mutations are mutually exclusive with TET2 and WT1 mutations, and are more commonly seen in cases with NPM1 and DNMT3A mutations [104].
Data are conflicting regarding the prognostic impact of IDH gene mutations. In one study, IDH2 mutations appeared to be associated with improved OS, but this benefit was only present in patients with IDH2 R140Q mutations [104]. Concurrent abnormalities in FLT3-ITD appeared to abrogate the beneficial impact of the IDH2 mutation. In another study, patients with IDH2 R172 mutations had a relatively good prognosis, similar to that seen with biallelic CEBPA mutations [18].
Mutant IDH1 can heterodimerize with wild-type IDH1 to create a mutant enzyme that converts alpha-ketoglutarate to 2-hydroxyglutarate (2-HG), which acts as an "oncometabolite" that blocks differentiation [145-148]. IDH1 mutations alone do not appear to be sufficient to induce transformation, but inhibition of mutant IDH1 appears to induce apoptosis and decrease replication in tumor models [149]. In one study, patients with IDH1/2 mutations had significantly increased serum 2-HG levels, and normalization of 2-HG levels after induction therapy was associated with better OS and DFS [150].
Enasidenib, an IDH2 inhibitor, is approved by the US Food and Drug Administration for treatment of patients with relapsed or refractory AML with mutant IDH2 [151]. Inhibitors of IDH1 (eg, AG120, IDH305) and both IDH1 and IDH2 (AG881) are undergoing clinical evaluation in AML and other hematologic malignancies. In particular, AG120 produced a high response rate in patients with relapsed or refractory AML with an IDH1 mutation and hence we also incorporate molecular analysis for IDH1 and IDH2 and NPM1 mutations into the pretreatment evaluation of all patients with newly diagnosed AML and normal karyotypes, regardless of age. (See "Treatment of relapsed or refractory acute myeloid leukemia", section on 'Remission re-induction'.)
KIT gene — Mutations of the KIT gene can be detected in approximately 6 percent of newly diagnosed AML and in 20 to 30 percent of patients with AML and either t(8;21) or inv(16) [104,152,153]. While some studies suggest that KIT gene mutations confer a higher risk of relapse and adversely affect OS in those with inv(16) [152,153], others suggest that this negative prognostic effect is only seen among AML with t(8;21) [104]. Screening for KIT mutations might also allow for use of tyrosine kinase inhibitors such as imatinib or dasatinib, which have in vitro activity against some (but not all) KIT mutations [154]. Clinical trials evaluating the addition of tyrosine kinase inhibitors in selected patients with KIT mutations are in progress.
In one study performed in older patients with high grade MDS and at least 20 percent of myeloblasts expressing KIT, treatment with a combination of imatinib and low dose cytosine arabinoside was unsuccessful [155]. The presence of KIT mutations was not assessed in this study.
WT1 gene — The Wilms tumor 1 gene (WT1) encodes a transcriptional regulator for genes involved in cellular growth and maturation. Disruption of this gene is thought to promote the proliferation of stem cells and disrupt cellular differentiation. Approximately 8 percent of AML cases and 13 percent of patients with cytogenetically normal AML will harbor mutations in WT1 [104,156]. Studies investigating the prognostic value of WT1 gene mutations or single nucleotide polymorphisms in cytogenetically normal AML have had mixed results [68,70,104,156-162]. Some report inferior rates of DFS and OS in patients with WT1 gene mutations while others do not.
ASXL1 and ASXL2 genes — The additional sex combs gene (ASXL1) is a human analog of the Drosophila gene located in chromosome 20q11. Mutations in the ASXL1 gene are present in 6 to 30 percent of cytogenetically normal AML and denote a poor prognosis [18,163-167]. The incidence of ASXL1 mutations in AML increases with age and is higher among those with a history of another myeloid malignancy (eg, myelodysplastic syndrome) [16,167,168].
ASXL1 mutations and NPM1 mutations are mutually exclusive, whereas ASXL1 mutations are strongly associated with alterations in regulators of RNA splicing, such as SRSF2 at 17q25 [18,164,169]. ASXL2 mutations are associated mutations of RUNX1 (also called AML1 or CBFA2) at 21q22 [18,169]. The prognostic impact of mutations in ASXL1 and SRSF2 appears to be additive; while the presence of either portends a poor prognosis, patients with both abnormalities have an even worse prognosis.
The biologic function of ASXL1 is unclear, but may be related to histone post-translational modifications.
DNMT3A gene — The DNMT3A (DNA [cytosine-5]-methyltransferase 3A) gene, located in 2p23.3, plays a role in epigenetic modifications necessary for mammalian development and cell differentiation. Mutations in DNMT3 lead to hypomethylation which, in turn, affects hematopoietic stem cell differentiation. Mutations in the DNMT3A gene are present in 20 to 22 percent of cytogenetically normal AML [79,80,85,86,88,91,170]. However, mutations of DNMT3A and other genes (eg, TET2, ASXL) are found in up to 10 percent of apparently normal adults 65 and older, increasing in incidence with advancing age, referred to as clonal hematopoiesis of indeterminate potential (CHIP) [171-173]. (See "Clonal hematopoiesis of indeterminate potential (CHIP) and related disorders of clonal hematopoiesis".)
Studies have reported mixed effects of DNMT3 mutations on prognosis. While DNMT3A mutations are generally associated with a poor prognosis, its prognostic impact may be affected by coexisting mutations in FLT3, NPM1, and IDH1/2. In one report, the presence of a DNMT3A mutation appeared to have a negative impact on the prognosis of patients without mutation in NPM1 or FLT3, but not of those with NPM1-mutated/FLT3 wild-type AML [91]. In contrast, another report found that DNMT3A mutations were associated with a worse outcome irrespective of the NPM1-mutation status [174]. The interpretation of DNMT3A mutation status is complicated by the high coincidence of DNMT3A mutations and NPM1 mutations (where the two mutations have opposite effects on prognosis) and a potential difference in effect according to the site of DNMT3A mutation.
TP53 gene — Mutations in the TP53 gene are seen in approximately 6 to 8 percent of de novo AML cases, often in cases with complex karyotype or other genetic abnormalities [18,175]. The prognostic impact of TP53 mutation and complex karyotype appears to be additive. Presence of either portends a poor prognosis, while patients with both abnormalities have an even worse prognosis.
Spliceosome mutations — Mutations of regulators of RNA splicing (eg, SRSF2, SF3B1, U2AF1, and ZRSR2) are often associated with mutations of chromatin-modifying genes (ASXL1, STAG2, BCOR, MLL, EZH2, and PHF6) and, together, patients with mutations in these two groups of genes constitute 18 percent of those with newly diagnosed AML [18]. Patients with these mutations are generally older, more frequently have antecedent myeloid disorders (eg, MDS), and have worse clinical outcomes.
Gene expression profiling — There is interest in the use of gene expression profiling (GEP) for the diagnosis, classification, and assessment of prognosis in AML, but it is not yet used routinely in clinical practice [176-186].
Several studies have analyzed leukemia cells from patients with AML and have identified gene "signatures" that may be used to distinguish subsets with different outcomes [177,187-190]. Subgroups with different gene expression profiles have been found in patients with normal cytogenetics [191], as well as those with well-defined cytogenetic changes, such as t(8;21) and inv(16) [192]. Other groups, such as t(15;17), appeared to be more homogeneous in their "signature."
GEP may also be able to identify other groups of AML patients with specific molecular signatures [178,193]. One cluster, associated with the lowest OS and highest cumulative relapse rate after CR, had a high frequency of poor prognostic markers (eg, del(7q), del(5q), t(9;22)). Normal CD34+ cells segregated into this cluster, suggesting that the molecular signature of treatment resistance resembles that of normal hematopoietic stem cells [194]. In another study, AML demonstrating a GEP signature of leukemic stem cells was associated with significantly worse OS, independent of other genetic features [195]. (See "Pathogenesis of acute myeloid leukemia", section on 'Transformation within primitive multipotent cells' and "Pathogenesis of acute myeloid leukemia", section on 'Leukemic stem cells'.)
There was still a wide range of outcomes in the prognostic groupings defined by GEP; accordingly, gene profiling in AML cannot as yet be used as a predictor in individual patients. However, a number of conclusions can be reached from these initial GEP data:
●They confirm the importance of cytogenetic subgroups of AML as relatively homogeneous diseases, since leukemias with distinct translocations tend to have very similar gene expression patterns [196].
●They begin to subdivide the large group of patients with normal karyotypes into different biological subsets that appear to have different outcomes [197].
●They support other data that suggest the role of a leukemic stem cell in the pathogenesis of AML [195].
●GEP studies may help to identify critical genes and their protein products whose expression could be modified by available drugs or new agents rationally designed to affect these targets [198].
●GEP may also help to define simpler algorithms incorporating a smaller number of genes, which might be detected using more widely available techniques, such as RT-PCR [176,199].
Before these findings can be applied clinically to define subclasses of patients, predict outcome, and perhaps even determine appropriate treatment, they need to be verified in larger groups of patients. In addition, the synthesis and clinical development of drugs to target these heterogeneous molecular changes will be very challenging.
MicroRNA expression profiling — MicroRNA are short sequences of single-stranded RNA that regulate gene expression. The role of microRNA expression profiling is becoming more prominent in our understanding of the pathogenesis of many cancers, including AML [67,200-208].
As an example, a small retrospective study that examined clinical outcomes in higher risk (FLT3 mutated, NPM wild-type) patients with cytogenetically normal AML reported that different patterns of microRNA expression are associated with varying rates of EFS [200]. While this was a small retrospective study that requires further confirmation, microRNA expression profiling analysis may lead to clues permitting treatment with agents selectively affecting specific microRNA targets in the future.
Another study in 187 younger adults (<60 years) with cytogenetically normal AML noted that patients whose tumors had higher expression of a single microRNA (miR-181a) had higher rates of CR and longer OS [209]. MiR-181a expression maintained its prognostic significance in tumors with FLT3-ITD and/or NPM1 wild-type. In contrast, another study of 363 patients with cytogenetically normal AML found that patients whose malignant cells had higher expression of another microRNA (miR-155) were less likely to attain a CR and had OS [210].
Further study is needed to determine how information regarding microRNA expression can be incorporated into clinical practice.
Tumor characteristics — Tumor characteristics can affect patient outcome. Characteristics that have been suggested as prognostic markers include the overexpression of drug efflux pumps, apoptosis inhibitors, antigen expression, or factors that lead to cell cycle progression [11,211-222]. It is unknown how best to integrate this information into clinical practice at this time. Examples follow.
MDR1 phenotype and P-glycoprotein — Overexpression of drug efflux pumps (eg, P-glycoprotein) has been associated with poor outcome in patients with AML, particularly in the elderly [211]. Conflicting observations about the effect of differences in multidrug resistance protein (MRP) expression have been published, perhaps related to the use of different assays. As an example, in one study detection of MRP1 by immunolabeling or RT-PCR failed to show a correlation with treatment outcome, whereas concomitant analysis via a functional assay did show a correlation [212].
The MDR1 phenotype may emerge during the evolutionary process of the leukemic cells. In a report from the SWOG, 71 percent of older patients with AML expressed MDR1, compared with 30 percent in younger subjects [11]. Patients who were MDR1 positive were less likely to have a CR and more likely to have resistant disease. The significance of this finding is unclear, since other studies have indicated that the presence of the MDR-1 phenotype is [213,214] or is not [215] associated with reduced OS in AML [216]. On the other hand, in the SWOG study, older patients who had MDR1-negative AML cells and favorable or intermediate cytogenetics had a high CR rate of 81 percent [11].
In a study involving 153 previously untreated patients with AML, positivity for P-glycoprotein (Pgp) did not adversely affect attainment of CR or OS unless Pgp was expressed along with lung resistance-related protein (LRP) [217]. The mean age of this latter sub-population (LRP+/Pgp+) was 64 years, whereas that of the other groups (LRP+/Pgp-, LRP-/Pgp+, and LRP-/Pgp-) was 48 years, indicating that this adverse prognostic combination was more common in the older patient.
There are a number of inhibitors of PgP mediated drug efflux that have been evaluated in randomized clinical trials. Unfortunately, these studies did not demonstrate improved outcomes in patients receiving the inhibitors in combination with chemotherapy and in some trials, toxicity was increased as well [223].
Overexpression of EVI1 — The ecotropic viral integration site 1 (EVI1) oncogene at chromosome 3q26 is abnormally overexpressed in approximately 8 to 10 percent of AML cases and appears to portend a poor prognosis [218,219,224]. EVI1 appears to interact with DNA methyltransferases that are involved in the epigenetic control of gene expression [225]. (See "Molecular genetics of acute myeloid leukemia", section on 'Mutations affecting DNA methylation'.)
As an example, a study of 1382 adults less than 60 years of age enrolled on multicenter prospective trials evaluated cases for EVI1 expression by polymerase chain reaction (PCR) [219]. EVI1 overexpression was detected in 38, 7, and 0.4 percent of cases with unfavorable, intermediate, and favorable risk cytogenetics, respectively. The prognostic effect of EVI1 overexpression was most apparent in the intermediate-risk group where overexpression was associated with a significantly lower rate of RFS and EFS and a nonsignificant trend toward decreased OS. (See "Cytogenetic abnormalities in acute myeloid leukemia".)
Overexpression of inhibitors of apoptosis — Overexpression of inhibitors of apoptosis, such as BCL2 [220] or survivin [221], may be associated with poorer outcome. High expression of the costimulatory molecule CD40 and the adhesion molecule CD11a have also been associated with a worse outcome [226]. In contrast, overexpression of the apoptosis promoter, BAX, or a high BAX/BCL2 ratio, was associated with better outcome in two studies [227,228]. Many studies are in progress evaluating the use of venetoclax, a BCL2 inhibitor, in AML [229].
CD25 expression — Expression of CD25 (IL-2 receptor alpha) may predict worse outcomes among patients with de novo AML. While initial small retrospective studies suggested that CD25 expression was associated with lower rates of CR and worse OS [230,231], its independent predictive value was questioned since CD25 expression has been shown to correlate with FLT3-ITD gene mutations [231].
The independent prognostic value of CD25 expression in patients with AML was best demonstrated in an analysis of 657 younger adults (≤60 years) with de novo AML treated in a prospective cooperative group trial (ECOG E1900) [232]. Expression of CD25 was identified in 87 patients (13 percent). When compared with CD25 negative tumors, those that express CD25 had inferior rates of CR and OS when stratified by cytogenetic risk group. Among the 75 patients with CD25 expression with mutational analysis results available, 57 (76 percent) demonstrated FLT3-ITD mutations. Among patients with FLT3-ITD mutations, expression of CD25 was associated with shorter median OS (10 versus 25 months) and a lower rate of OS at three years (4 versus 42 percent).
MEASURABLE RESIDUAL DISEASE (MRD) — Measurable residual disease (MRD; also referred to as minimal residual disease) refers to detection of malignant cells, even in the setting of apparent hematologic complete remission. The prognostic value of MRD for AML is presently uncertain. (See "Induction therapy for acute myeloid leukemia in medically-fit adults", section on 'Remission assessment'.)
Methods for detecting MRD in AML include multiparameter flow cytometry, polymerase chain reaction (PCR), and next-generation sequencing. Details of MRD techniques, standardization of assays, types and timing of samples, and other technical aspects of MRD detection are described separately. (See "Induction therapy for acute myeloid leukemia in medically-fit adults".)
Studies that have evaluated the prognostic value of MRD in AML are discussed separately. (See "Induction therapy for acute myeloid leukemia in medically-fit adults", section on 'Remission assessment'.)
INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5th to 6th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10th to 12th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.
Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient education" and the keyword(s) of interest.)
●Beyond the Basics topics (see "Patient education: Acute myeloid leukemia (AML) treatment in adults (Beyond the Basics)")
SUMMARY
●The term acute myeloid leukemia (AML) refers to a group of hematopoietic neoplasms involving cells committed to the myeloid line of cellular development. AML is characterized by a clonal proliferation of myeloid precursors with reduced capacity to differentiate into more mature cellular elements.
●The response to treatment and overall survival of patients with AML is heterogeneous. A number of prognostic factors related to patient and tumor characteristics have been described for AML (table 1). Of these, patient age at diagnosis, performance status, and karyotype have the most direct effect on treatment at this time and should be a part of the initial evaluation of all patients with newly diagnosed AML. (See 'Clinical risk factors' above and 'Karyotype' above.)
●The clinical role of gene mutation analysis, gene expression profiling, and microRNA profiling remains uncertain at this time, although a number of mutations and changes in levels of certain proteins have prognostic impact and increasingly are part of the "routine" characterization of AML (table 4 and table 5 and table 6).